Last.fm – the Blog

20 DECEMBER 2007 UPDATE:
Hello world! We’ve now extended our YouTube integration beta to most countries round the globe. There’s also a new ‘flag as incorrect’ feature that you can use to help fix up pages where we’ve accidentally imported the wrong videos. Thanks for all your feedback so far! -Matt

—

Why let mashups have all the fun?

Today we’re testing a new feature for users on our home turf: music videos from YouTube. These videos sit alongside our own videos, and are automatically connected when you visit track pages.

If you’re in the UK, check out the pages for your favourite tracks, and let us know what you think! Depending on the response, we may consider rolling this out to other parts of the world too.

There were people missing from the Last.fm office on Monday. Some didn’t survive the moshpit that erupted as SSS brought their flashbomb thrash to our second ever live event at the Old Blue Last on Saturday. We would mourn, but quite frankly we’re still reeling from an evening of metal mayhem, sci fi drum madness and punishing electro-noise.

We took over the Vice pub in East London in the early evening on Saturday, and had to pull the plug in the early hours of Sunday when the police descended to tell us off for being too loud. (They should’ve been upstairs when Errorplains kicked off with a wall of FX-pedal noise. Now that was loud.)

In between the smash’n’grab metal of SSS and the synth assault of The Errorplains , crazy drum maestro Duracell played a rare UK live set. If you haven’t seen this guy before, he manages to produce entire sets of future-music using just a drum set, trigger machine and a laptop. Never has a man been so sweaty as Duracell was by the time his incredible one-man beat-symphony came to an end.

Downstairs was equally rammed, as punters wigged out to our very own Last.fm DJs, and the deck-wrecking moves of Tape Deck, Shitting Fists and People Are Germs. By the end of the night the walls were dripping and bouncers were holding back a multitude outside.

A ton of thanks goes to the performers of the night, SSS, Duracell, The Errorplains, The Tapedeck DJ’s, Shitting Fists, People are Germs and our very own Last.fm DJ team. And last but not least, to all of the people that came down to party along side us in between the sweaty walls of the Old Blue Last, we salute you!

For all those who missed out on all the fun and nuttery of Saturday night you can check it out here, but to really get a feel for it come check out our upcoming shindig we’re throwing on November 7th to kick off the UK tour for The Acid Mothers Temple & The Melting Paraiso U.F.O.. – A nice little way to follow up our rock-out sauna party. ;-)

What a week. From the moment Radiohead announced their plans for latest release ‘In Rainbows’ we’ve been itching to see how you, the listeners, would respond. We let our friends over at Drowned In Sound in on some early data after just 12 hours. Now that our weekly charts are compiled there’s not much else to say apart from “Wowzer”. Below you can see the Last.fm global tracks chart for last week, and below that the official UK downloads chart for the same period. Spot the difference?

Last.fm global tracks chart for week ending October 14th

Official UK downloads chart – Week Starting October 15th

It’s more likely than you think…

Here at Last.HQ, we get a wonderful weekly delivery of fruit, which seemingly takes about 4 weeks to ripen. That aside however, there are oft good things to be found in the boxes that arrive. They ship a little sheet along with the fruit too; though, we haven’t worked out yet whether this is to aide in the identification of said fruit, or just to subversively teach us that fruit is good.

While perusing the sheet of fruity goodness, we suddenly happened upon a here-to-undiscovered delight, and quickly set to testing it in our lab… The ClemePod.

Immediately, we thought that perhaps Apple had released a new Pod device, in the shape of a fruit, and had secretly shipped us a test batch to add experimental scrobbling support to. We grabbed some speakers and plugged it in to see if, despite the lack of user interface development, it would play…

To our delight, a hum came forth from the speakers. Unfortunately, as yet, our client team have been unable to decipher the interface, so scrobbling is likely to be impossible for the moment.

Watch this space for more ClemePod device updates.

The Last.fm Research group links to an interesting article on the Hometracked blog: All Linkin Park songs look the same

In the author’s words:

Each image below shows the audio level in (roughly) the first 90 seconds of a Linkin Park song. Note that I adjusted the tempo of a few tracks for better visual alignment.

Make sure to read the full article to see more pics and for the full explanation.

Do you disagree with the implications? Does this say anything about the state of Pop? What other patterns could one find in different forms of pop music? Discuss. (But pls stay civil :)

Oh and — has anyone done similar analyses for other artists as well? Let us know if you have something to show.

Update: now giving credit where credit is due… Thanks aradnuk for pointing out the error.

This Tuesday brings some piping hot new features from the webteam oven! Here’s what we snuck onto the site today.

Recommended music videos on your dashboard

A page full of videos from artists you’ve listened to, with newly-uploaded stuff at the top so you always get the freshest content. Finally, eh?

Larger video player

If you get enough emails promising that “bigger is better,” you start to believe it.

Check out our enlarged player in action:

• Halloween comes early courtesy of the inimitable Messer Chups

• Can you find a fiercer drummer than Last.fm’s own Matt Hillman? (I’m afraid to ask.)

• Explore the softer side of Last.fm video with Fink.

Portable Taste-o-meter

Embed our new tasteometer widget and anyone who wanders past your Myspace or blog can compare music taste with you. It won’t be as good as yours, of course, but whose is really. It’s compare…o…licious…ness.

Until next time, I leave you with this photo of a Last.fm Toadstool, courtesy Wesley Mason’s Flickr stream.

Researchers all over the world have started using Last.fm tags (which are available through our open API) in their studies. So the next time you tag something cheese on toast or woopwoop you might actually cause some researchers somewhere out there sleepless nights while they are trying to understand what kind of music those tags define ;-)

Here are some research papers that were presented at the International Conference on Music Information Retrieval in Vienna last week and that used Last.fm tags:

• Eck, Bertin-Mahieux, & Lamere (USA), Autotagging music using supervised machine learning: In this paper the authors describe their work on algorithms that try to tag music like humans. I saw a demo and the results are impressive (but I doubt they’ll ever get woopwoop and related tags right). To learn more about their work check out Lamere’s blog.

• Geleijnse, Schedl, & Knees (Netherlands & Austria), The quest for ground truth in musical artist tagging in the social web era: The authors use tags to compute the similarity of artists and classify them into genres. Geleijnse also made some excellent points on why tags used by a community of listeners are so interesting (compared to categories invented by experts who might not even like the music they are talking about).

• Hu & Downie (USA), Exploring mood metadata: Relationships with genre, artist and usage metadata: The authors use Last.fm tags to gain more insight into how mood tags are linked to other tags (e.g. “music for exercising”) and artists.

• Hu, Mert, & Downie (USA), Creating a simplified music mood classification ground-truth set: The authors used Last.fm tags to set up test sets which are used to improve algorithms that can classify moods.

• Levy & Sandler (UK), A semantic space for music derived from social tags: The authors use tags to compute music similarity and investigate important dimensions of similarity. It’s nice to see our neighbours in East London doing such interesting work :-)

We’ve dumped millions of fingerprints into our new-and-improved sausage grinder over the past few days. We’re carefully sifting through the results, examining a lot of anomalies and edge cases. Here are some interesting ones:

• You can listen to 331 different renditions of Happy Birthday by Ingrid DuMosch on our site. They differ only in the addressee of the song. For me, Happy Birthday BUBBA is the clear winner.
• Canadian band Barenaked Ladies has released at least forty recordings of different live concerts that all contain exactly the same comedy routine! Way to mash play on that banter box, fellas.
• Clan of Xymox (favorite band of one of our very own) released a live version of their song Consolation – the two songs differ only via some compression, maybe a little reverb, and some cheering at the beginning and end. What this means is an exercise left to the reader.
• An underlying notion in fingerprinting is to recognize a song by what makes it unique. So songs such as this or this present a challenge. Similarly, what if your CD-ripper automatically cuts your copy of John Cage’s 4’33” down to 4’29”? Tough luck, buddy.
• Burgler Alarm!

But we’ve been doing more than just listening to music. In fact, we just released a new similar tracks feature on the site. Here are the similar tracks to Smog’s Our Anniversary, a recent favorite of mine.

Back in the early zeroes, two of the three founders of Last.fm used to put on events at The Spitz, an East London venue that’s been showcasing new bands almost as long as Spitalfields has had a market (which is 369 years, fact fans). Then they got distracted by all this web start-up malarkey.

But yesterday they returned – with the rest of us, plus five amazing bands, our finest Last.fm deck-wreckers, and special guest DJs Hot Chip, who we managed to coax away from the studio where they’re finishing off their third album. Altogether, we rocked the place.

Upstairs was a riot of guitar and synth-fuelled mayhem as five of the finest new bands known to humanity (including Everyone To The Anderson, with our very own Matt Hillman on drums) did their thing to an audience of hardcore Last.fm users, music freaks, and the entirety of our office – including a moshing founder or two.

Downstairs, our one-woman Queen of Noize Fiona and Last.fm’s Spanish mixmaster Miguel kept the tunes flowing all night – from an amazing playlist drawn from the scrobble stats of all the Last.fm users attending the party (and there were quite a lot of you guys by the end!) Then Hot Chip’s Joe Goddard took over and slammed down an hour of genre-mashing tune-age, including two new tracks from the Chip’s next album – ‘Ready For The Floor’ and ‘Shake A Fist’ – that totally blew up the room.

(See more photos on Flickr)

By the end of the night we’d packed the place to the rafters, drank all cold beer in the house (!), and raised a ton of money to help the Spitz find a new home. Not bad for our first ever live event.

The audio from last night now live on the group page, and we’ll be posting the video soon so hold tight for more updates! Join the Last.fm Presents group to keep up with future events.

Thanks goes out to Turbowolf, Lost Penguin, Mica Levi and The Cluster, Everyone To the Anderson, and Agaskodo Teliverek for providing some of the best live music we’ve all seen in ages! And if you’re in London and want to throw The Spitz some more love, check out their One Day Punk Blues Festival on Sunday.

The previous post about beta testing our new fingerprint technology generated quite a bit of feedback – thanks to everyone involved.

Quick Update

• 10 million fingerprint submissions received so far (keep ’em coming)
• Lots of useful feedback and thoughtful questions
• We are frantically working on the server architecture in order to get a public-facing lookup service ready as soon as possible.
• Dedicated hardware for this is scheduled to arrive on Friday afternoon. Laurie will probably post pictures once all the LEDs are working.

Gory details

This weekend we started feeding the fingerprints en-mass into the fingerprint indexing service, which will tell us how many duplicate tracks there are and so on. This is in full swing, but we’ve only imported about 3m so far. A graph showing total fingerprints received, and total unique tracks is available here (updates twice an hour) – bear in mind that we have not imported all the fingerprints yet, so expect the ratio of ‘unique tracks’ / ‘fingerprints received’ to increase over the course of the week.

Although far from complete, we have the capability to query the currently imported data and return an internal unique ID, and the list of spellings used for that fingerprint:

$ ./query ../mp3/Pink\ Floyd/The\ wall\ \(disc\ 1\)/03\ Another\ brick\ in\ the\ wall\ \(part\ 1\).mp3
Fingerprint ID: 527153
--
Pink Floyd - Another Brick in the Wall (Part 1)
Pink Floyd - Another Brick In The Wall (Part I)
Pink Floyd - Another Brick In The Wall Part 1
Pink Floyd - Another Brick in the Wall, Par
Pink Floyd - Another Brick in the Wall, Pt. 1
Pink Floyd - The Wall (PartI)
Pink Floyd - Another brick in the wall (p..
Unknown - unknown

Or another example:

$ ./query ../mp3/Guns\ N\'\ Roses/Use\ Your\ Illusion\ II/04\ -\ Knockin\'\ On\ Heaven\'s\ Door.mp3
Fingerprint ID: 1211395
--
Guns N' Roses - Knockin' on Heaven's Door
Guns N' Roses - Knockin´ On Heaven´s Door
Guns N' Roses - Knockin On Heavens Door

I’m expecting a lot more versions of this one once we’ve cleared the fingerprint backlog… all those apostrophes ;)

What this shows is simply the list of known spellings for songs that sound exactly the same as the MP3s used as a test. The list of known spellings will increase as we import the rest of the data. It’s going to be trivial for us to order this list by popularity, and assume that the most popular spelling is usually correct. We are however, painfully aware that this is not going to be reliable enough in many cases.

The crux of the moderation task will be taking these sets of spellings (once album and popularity data is added) and distilling it into a single correct version

Once we have processed the fingerprint backlog (and hopefully updated the client to address some known issues) we will be in a better position to figure out how we’re going to do this.

Initial questions we hope to answer:

1. How accurate will ‘the most popular spelling is correct’ assumption be?
2. How many unique recordings make up the ~10m fingerprint submissions?
3. Can we work with MusicBrainz for voting, and to publish a “common misspellings” dataset?
4. Just how many ways to write “Guns N’ Roses – Knockin’ on Heaven’s Door” are there?